Interictal EEG Signals Research Articles

Seizure detection using nonlinear measures over EEG frequency bands and deep learning classifiers

Epilepsy is a brain disorder that causes patients to suffer from convulsions, which affects their behavior and way of life. Epilepsy can be detected with electroencephalograms (EEGs), which record brain neural activity. Traditional approaches for detecting epileptic seizures from an EEG signal are time-consuming and annoying. To supersede these traditional methods, a myriad of automated seizure detection frameworks based on machine learning techniques have recently been developed. Feature extraction and classification are the two essential phases for seizure detection. The classifier assigns the proper class label after feature extraction lowers the input pattern space while maintaining useful features. This paper proposes a new feature extraction method based on calculating nonlinear features from the most relevant EEG frequency bands. The EEG signal is first decomposed into smaller time segments from which a vector of nonlinear features is computed and supplied to machine learning (ML) and deep learning (DL) classifiers. Experiments on the Bonn dataset reveals an accuracy of 99.7% reached in classifying normal and ictal EEG signals; and an accuracy of 98.8% in the discrimination of ictal and interictal EEG signals. Furthermore, a performance of 100% is achieved on the Hauz Khas dataset. The classification results of the proposed approach were compared to those from published state of the art techniques. Our results are equivalent to or better than some recent studies appeared in the literature.

EEG based automated detection of seizure using machine learning approach and traditional features

The detection of epileptic seizures is key for neurologists to initiate the right treatment at the earliest. However, the traditional methods are dependent on manual diagnosis which are error-prone. Hence, the current article presents an automated machine learning-based approach with competing results to detect seizures and its types using traditional features. Towards this goal, the present paper explores the performance of two fractal parameters namely Higuchi and, Kat’z fractal dimension, along with power spectral density and spectral entropy to detect EEG signals with seizure and non-seizure characteristics. To validate the performance via cross-validation technique, the features are extracted and experimented from three publically available databases such as Bern-Barcelona (focal, non-focal), Khas (preictal, interictal, ictal), and Bonn. Thereafter evaluated with different machine learning algorithms like Random Forest classifier, Ada Boost classifier, Gradient Boosting classifier, Extra Tree classifier, SVM classifier, and Naive Bayes classifier. The experiments showed 100% F1 score and accuracy in classifying focal and non-focal signals for random forest and extra tree classifier methods. To the best of our knowledge, this is the first paper reporting this score for the Bern-Barcelona dataset. The same features are also experimented for the detection of interictal, ictal and preictal scenarios. In the process, 100% accuracy is obtained in classifying interictal and ictal EEG signals, and an accuracy of 94% is attained in classifying interictal-preictal signals. This score is 14% higher than the current state-of-the-art methods using the same database. The features are also used in the Bonn database, where the proposed approach gave an accuracy of 100% same as the state-of-the-art methods.

Convolutional neural network-based lightweight hardware IP core design for EEG epilepsy prediction

Epilepsy is one of the most common neurological disorders. In order to realize a portable and wearable epilepsy prediction device, a lightweight hardware IP core is proposed for EEG epilepsy prediction. The scheme first analyzes the epileptic EEG signal and extracts the time-frequency feature maps of interictal and preictal EEG signals. Then it reduces the network parameters using the single-channel approach to reduce the number of neurons and network complexity, and trains the lightweight epilepsy prediction convolutional neural network by combining the time-frequency feature maps. Finally, a hardware convolutional layer is designed with pipeline function, and it connect 8 layers with feature map buffer to realizes the lightweight EEG epilepsy prediction hardware IP core. Using TSMC 65 nm process, the hardware IP core is functionally verified with 87.9% prediction accuracy. The area is 2.55 mm2 and the power is 4.45 mW. Under 1V & 20 MHz operating conditions, single prediction latency is 3.29 ms and energy efficiency is 0.146 μJ/class.

Epileptic seizure detection on a compressed EEG signal using energy measurement

Epilepsy is a common neurological disorder affecting both children and adults. It can trigger seizures without any stimuli. An accurate diagnosis of epilepsy is essential to the treatment and management of the patient’s condition. Electroencephalography EEG is a non-invasive technique for the diagnosis of epilepsy. Data analysis and processing can be challenging due to the high dimensionality of EEG signals. In this study, the dimensionality reduction technique using compressive sensing is applied to EEG signals to optimize the process and improve the accuracy of the classification in the epileptic EEG signals. The CS method is applied to minimize processed data by reducing EEG signal dimensions, and energy is calculated on the compressed signal as a feature. The performance of the proposed method is evaluated using SVM with Linear, Quadratic, Cubic, and Radian Basis Function kernels through 5- and 10-fold cross-validation. The results showed 100% accuracy in the classification of ictal and inter-ictal EEG signals with compression ratios of 1/2, 1/4, and 1/8. The results demonstrate that CS is effective in analyzing EEG and diagnosing epilepsy. In addition, this study has significant implications for further development of dimensionality reduction techniques in various EEG-based applications, such as epilepsy classification, sleep stage analysis, and brain–computer interface.

Automated detection of epileptic seizures using multiscale and refined composite multiscale dispersion entropy

Epilepsy is one of the most common neurological disorders. The electroencephalogram (EEG) is a valuable tool for the detection of epileptic seizures. The diagnosis of epilepsy requires the neurologists to continuously monitor the long-term EEG recordings of patients, which is a time-consuming and error-prone procedure. Therefore, automatic epileptic seizure detection becomes essential. Entropy-based methods are widely used for the automated detection of seizures from EEG signals due to the nonlinear and chaotic nature of these signals. In this work, we propose two recently introduced entropy features, multiscale dispersion entropy (MDE) and refined composite multiscale dispersion entropy (RCMDE) for detection of seizures. We assess the ability of MDE and RCMDE to discriminate the normal EEGs of healthy subjects, interictal (in between seizures), and ictal (during seizures) EEGs of epilepsy patients. We also investigate the two parameters namely, number of classes c and embedding dimension m of MDE and RCMDE that provide the best performance for seizure detection. For this purpose, the MDE and RCMDE values are estimated from normal, interictal, and ictal EEG signals, one-way ANOVA test is employed, and significant features are fed to a support vector machine (SVM) classifier. The experimental results demonstrate that both MDE and RCMDE are promising feature extraction methods that can quantify the complexity of EEG signals successfully and the highest classification accuracies were obtained when c=5 and m=3 for both MDE and RCMDE frameworks. Besides, we have compared the proposed MDE and RCMDE classification results with that of multiscale entropy (MSE) and multiscale permutation entropy (MPE) methods that have been previously applied for the study of seizure detection. It was found that MDE and RCMDE contribute significantly in improving the accuracy of seizure detection.

Wavelet Time Scattering Based Classification of Interictal and Preictal EEG Signals

If it were possible to reliably identify the preictal brain state from dynamical changes in EEG data of epilepsy patients, then the age long problem of actualizing a fully automated closed-loop seizure – warning or seizure-prevention system that is clinically deployable would have been resolved. Accordingly, through feature engineering, a great deal of effort has been invested over the discovery of EEG features or measures that are always indicative of the preictal brain state. However, this has proven to be difficult, time consuming and apparently unsuccessful. Therefore, lately, attention has shifted to feature learning-methods that automatically learn and extracts useful discriminatory features from raw data. This paper studies the efficacy of wavelet time scattering learned EEG features for interictal and preictal EEG classification. Wavelet time scattering network developed in Matlab and two different EEG datasets: CHB-MIT scalp EEG and AES intracranial EEG datasets were used for the study. The learned interictal and preictal EEG features were used to train and evaluate a simple binary support vector machine classifier. Three different classification accuracy results namely ordinary cross validation, true cross validation and test classification accuracy results were reported for the analysis. Mean classification accuracy values of 93.15%, 97.57% and 91.33% were obtained respectively for the scalp EEG while mean classification accuracy values of 98.33%, 100% and 96.73% were obtained respectively for the intracranial EEG. A general comparison showed that the combination of wavelet time scattering learned EEG features and a simple binary support vector machine classifier performed equally or even better than deep convolutional neural networks in EEG classification tasks. Finally, wavelet time scattering has proven to be a very good EEG feature learner and may greatly improve the sensitivity and specificity of seizure prediction algorithms.

Assessing multi-layered nonlinear characteristics of ECG/EEG signal via adaptive kernel density estimation-based hierarchical entropies

Background and objectiveCoarse-grained analysis-based entropies are capable of quantifying the nonlinear characteristic of signals in different scales but they ignore the high-frequency information. We here proposed hierarchical entropies via which the low- and high-frequency information of a signal can be both revealed so that better understanding of signal’s dynamic characteristics can be reached. MethodsIn this paper, adaptive kernel density estimation was used to estimate probability densities for distribution entropy (DistEn), fuzzy DistEn (FDistEn), complex-valued DistEn (CVFDistEn) and complex-valued FDistEn (CVFDistEn). Totally six hierarchical entropies were then raised, to overcome the defects of traditional coarse-grained operation and to assess the complexity of the full-band components of a signal. Fusion methods of distribution entropy-derived hierarchical entropies and probabilistic neural network (PNN) were finally put forward to identify normal and congestive heart failure (CHF) electrocardiogram (ECG) as well as normal, interictal and ictal electroencephalography (EEG) signals. ResultsExperimental results indicate the proposed hierarchical entropies can characterize the complexity of ECG and EEG signals in different scales. Moreover, fusion methodology of hierarchical FDistEn and PNN achieved the highest mean Matthews correlation coefficient of 100 % in distinguishing normal and CHF ECG signals, while combination of hierarchical CVFDistEn and PNN reported the best mean accuracy of 99.23 ± 0.23 % for identification of normal, interictal and ictal EEG signals. ConclusionsOur proposed adaptive kernel density estimation-based hierarchical entropies can characterize ECG and EEG signals effectively. The fusion methods of hierarchical entropies and PNN bring a new tool for identification of different types of ECG and EEG signals.

EEG signal classification method based on improved empirical mode decomposition and SVM

Epilepsy is a common phenomenon formed by abnormal discharges between brain neurons. The seizures of epilepsy are sudden and irregular. As a non-stationary signal, EEG signals can express its characteristics to a certain extent, and makes a significant difference in the monitoring and treatment of epilepsy diseases. This study employs empirical mode decomposition (EMD) to decompose the interictal and epileptic EEG signals into multiple eigenmode functions (IMF), and combines the correlation coefficient to screen the main IMF and extract its variance, fluctuation coefficient and Coefficient of variation and other features, combined with support vector machines for classification. Compared with the traditional empirical mode decomposition, this method has higher accuracy in the identification and classification of epileptic signals. The combination of this method not only provides a theoretical method for disease diagnosis and treatment, but also verifies the research and application value of EEG signals to a certain extent.

Automatic recognition of preictal and interictal EEG signals using 1D-capsule networks

Epilepsy is the most common neurological disorder affecting people of all ages. Seizure prediction can be achieved by separating the preictal state in which the changes in the brain activities begin to occur from the interictal state. Therefore, in this study, a novel method for preictal/interictal recognition, the most important step in seizure prediction from scalp electroencephalogram signals, is proposed. In the proposed method, one-dimensional capsule networks, a novel neural network model, is used. The best classification accuracy for preictal/interictal recognition was achieved with 97.74% in F3-C3 channel pairs. Compared to other methods, our 1D-CapsNet model achieved the best performance. Moreover, the results indicated that the interval that ended 30 min before the onset of seizures contained important information about preictal/interictal recognition. We believe that the proposed method will bring a new perspective to the seizure prediction studies of capsule networks.

Epileptogenic neurophysiological feature analysis based on an improved neural mass model

To elucidate the neurophysiological mechanisms underlying seizures according to electroencephalogram (EEG) signals, a neurophysiologically-based neural mass model that can produce EEG-like signals was adopted to simulate ictal and interictal EEG signals. A delay unit and a gain unit were added to the Wendling model to fit EEG signals in the time domain. An optimal parameter set minimising the error between observed and simulated EEG was identified using a genetic algorithm. To compare the inhibition and excitation during the ictal and interictal periods, the model parameters were determined for two sets of EEG signals using the proposed method. The results show that the model with identified parameters can simulate the real EEG signal well, with a mean square error of 0.0315-0.2138. Fifty repetitions for every selected EEG signal showed that the dispersion of the identified parameters was small in most cases, and the identification procedure generally showed similar values. Comparison of the model parameters of seizure and non-seizure EEG signals showed enhanced excitability, attenuated inhibition, and a more concentrated energy distribution in the frequency domain during the ictal periods. The experimental results for long-term EEG signals revealed continuous changes in the model parameters during epileptic seizures.

Epileptogenic neurophysiological feature analysis based on an improved neural mass model

To elucidate the neurophysiological mechanisms underlying seizures according to electroencephalogram (EEG) signals, a neurophysiologically-based neural mass model that can produce EEG-like signals was adopted to simulate ictal and interictal EEG signals. A delay unit and a gain unit were added to the Wendling model to fit EEG signals in the time domain. An optimal parameter set minimising the error between observed and simulated EEG was identified using a genetic algorithm. To compare the inhibition and excitation during the ictal and interictal periods, the model parameters were determined for two sets of EEG signals using the proposed method. The results show that the model with identified parameters can simulate the real EEG signal well, with a mean square error of 0.0315-0.2138. Fifty repetitions for every selected EEG signal showed that the dispersion of the identified parameters was small in most cases, and the identification procedure generally showed similar values. Comparison of the model parameters of seizure and non-seizure EEG signals showed enhanced excitability, attenuated inhibition, and a more concentrated energy distribution in the frequency domain during the ictal periods. The experimental results for long-term EEG signals revealed continuous changes in the model parameters during epileptic seizures.

Integrating old and new complexity measures toward automated seizure detection from long-term video EEG recordings

SummaryAutomated seizure detection in long-term video-EEG recordings is far from being integrated into common clinical practice. Here, we leverage classical and state-of-the-art complexity measures to robustly and automatically detect seizures from scalp recordings. Brain activity is scored through eight features, encompassing traditional time domain and novel measures of recurrence. A binary classification algorithm tailored to treat unbalanced dataset is used to determine whether a time window is ictal or non-ictal from its features. The application of the algorithm on a cohort of ten adult patients with focal refractory epilepsy indicates sensitivity, specificity, and accuracy of 90%, along with a true alarm rate of 95% and less than four false alarms per day. The proposed approach emphasizes ictal patterns against noisy background without the need of data preprocessing. Finally, we benchmark our approach against previous studies on two publicly available datasets, demonstrating the good performance of our algorithm.

A Feature Tensor-Based Epileptic Detection Model Based on Improved Edge Removal Approach for Directed Brain Networks.

Electroencephalograph (EEG) plays a significant role in the diagnostics process of epilepsy, but the detection rate is unsatisfactory when the length of interictal EEG signals is relatively short. Although the deliberate attacking theories for undirected brain network based on node removal method can extract potential network features, the node removal method fails to sufficiently consider the directionality of brain electrical activities. To solve the problems above, this study proposes a feature tensor-based epileptic detection method of directed brain networks. First, a directed functional brain network is constructed by calculating the transfer entropy of EEG signals between different electrodes. Second, the edge removal method is used to imitate the disruptions of brain connectivity, which may be related to the disorder of brain diseases, to obtain a sequence of residual networks. After that, topological features of these residual networks are extracted based on graph theory for constructing a five-way feature tensor. To exploit the inherent interactions among multiple modes of the feature tensor, this study uses the Tucker decomposition method to get a core tensor which is finally reshaped into a vector and input into the support vectors machine (SVM) classifier. Experiment results suggest that the proposed method has better epileptic screening performance for short-term interictal EEG data.

A Novel Geometrical Method for Discrimination of Normal, Interictal and Ictal EEG Signals

A Novel Geometrical Method for Discrimination of Normal, Interictal and Ictal EEG Signals

TTA, a new approach to estimate Hurst exponent with less estimation error and computational time

Investigation of long memory processes in signals can give us an important information about how signals have behaved so far and how will it behave in future. Hurst exponent estimation is a proper tool to show memory in signals. Rescaled range analysis (R/S), detrended fluctuation analysis (DFA) and generalized Hurst exponent (GHE) are most known methods for estimation of Hurst exponent which introduced in literature. In this paper, we propose a new algorithm to estimate Hurst exponent based on triangles total areas (TTA) that can be made out of three samples of different lag in time series. To test our algorithm performance, we used two kinds of synthetic waveforms with known Hurst exponents. Results indicates that the proposed method is superior with respect to data length, estimation error, computational time and noise sensitivity. We also apply our proposed method in epilepsy detection and compare our results with previous works to show outperformance of our algorithm with accuracy of 94.5% in classification between interictal and ictal EEG signals.

Identification of epileptic seizures in EEG signals using time-scale decomposition (ITD), discrete wavelet transform (DWT), phase space reconstruction (PSR) and neural networks

Traditionally, detection of epileptic seizures based on the visual inspection of neurologists is tedious, laborious and subjective. To overcome such disadvantages, numerous seizure detection techniques involving signal processing and machine learning tools have been developed. However, there still remain the problems of automatic detection with high efficiency and accuracy in distinguishing normal, interictal and ictal electroencephalogram (EEG) signals. In this study we propose a novel method for automatic identification of epileptic seizures in singe-channel EEG signals based upon time-scale decomposition (ITD), discrete wavelet transform (DWT), phase space reconstruction (PSR) and neural networks. First, EEG signals are decomposed into a series of proper rotation components (PRCs) and a baseline signal by using the ITD method. The first two PRCs of the EEG signals are extracted, which contain most of the EEG signals’ energy and are considered to be the predominant PRCs. Second, four levels DWT is employed to decompose the predominant PRCs into different frequency bands, in which third-order Daubechies (db3) wavelet function is selected for analysis. Third, phase space of the PRCs is reconstructed based on db3, in which the properties associated with the nonlinear EEG system dynamics are preserved. Three-dimensional (3D) PSR together with Euclidean distance (ED) has been utilized to derive features, which demonstrate significant difference in EEG system dynamics between normal, interictal and ictal EEG signals. Fourth, neural networks are then used to model, identify and classify EEG system dynamics between normal (healthy), interictal and ictal EEG signals. Finally, experiments are carried out on the University of Bonn’s widely used and publicly available epilepsy dataset to assess the effectiveness of the proposed method. By using the 10-fold cross-validation style, the achieved average classification accuracy for eleven cases is reported to be 98.15%. Compared with many state-of-the-art methods, the results demonstrate superior performance and the proposed method can serve as a potential candidate for the automatic detection of seizure EEG signals in the clinical application.

Cross-frequency network analysis of functional brain connectivity in temporal lobe epilepsy**Project supported by the National Natural Science Foundation of China (Grant No. 61302002).

In this study, we investigate the cross-frequency coupling and functional brain networks in the subjects with temporal lobe epilepsy (TLE) using interictal EEG signals. The phase to phase synchronization within and across frequency bands is calculated and a significant difference between the epilepsy and control groups is observed. Compared with the controls, the epilepsy patients exhibit a stronger within-frequency coupling (WFC) within theta and beta bands, and shows a stronger cross-frequency coupling (CFC) in the delta–alpha and theta–alpha band pairs, but a weakened CFC in alpha–beta band pairs. The weakened coupling between alpha and high frequency band reflects a suppression of phase modulation between the brain regions related to epilepsy. Moreover, WFC and CFC are positively correlated, which is higher in the patients relative to controls. We further reconstruct functional brain connectivity and find that both WFC and CFC networks show small-world properties. For the epilepsy, the small-world efficiency is enhanced in the CFC networks in delta–alpha and theta–alpha band pairs, whereas weakened between alpha and beta bands, which suggests a shift away from the optimal operating point in the epileptic brain with a new balance between WFC and CFC. Our results may help us to understand the important role of information communication across different frequency bands and shed new light on the study of pathology of epilepsy.

Electroclinical characteristics of seizures arising from the precuneus based on stereoelectroencephalography (SEEG)

BackgroundSeizures arising from the precuneus are rare, and few studies have aimed at characterizing the clinical presentation of such seizures within the anatomic context of the frontoparietal circuits. We aimed to characterize the electrophysiological properties and clinical features of seizures arising from the precuneus based on data from stereoelectroencephalography (SEEG).MethodsThe present retrospective study included 10 patients with medically intractable epilepsy, all of whom were diagnosed with precuneal epilepsy via stereoelectroencephalography (SEEG) at Yuquan Hospital and Xuan Wu Hospital between 2014 and 2016. Clinical semiology, scalp electroencephalography (EEG) findings, magnetic resonance images (MRI), and positron emission tomography (PET) images were analyzed during phase I preoperative evaluations. Following electrode implantation, the semiological sequence, ictal SEEG evolution, and anatomy of the relevant brain structures were analyzed for each seizure.ResultsSeven of ten patients reported auras, including body image disturbance (2/7), vestibular responses (2/7), somatosensory auras (1/7), visual auras (1/7), and non-specific auras (1/7). Primary motor manifestations included bilateral asymmetric tonic seizures (BATS) (7/10) and hypermotor seizures (HMS) (3/10). In one patient, epileptiform discharge on interictal EEG occurred ipsilateral to the side of the epileptogenic zone (EZ). Discharge was non-lateralized in the remaining nine patients. In six patients, interictal EEG signals were primarily localized in the temporal–parietal–occipital area. In two patients, ictal onset occurred ipsilateral to the EZ, which was mainly located in the temporal–parietal–occipital area. Two patterns of seizure spread were observed. The first pattern was characterized by BATS activity with ictal spread to the supplementary motor area (SMA), paracentral lobule (PCL), precentral gyrus (PrCG), or postcentral gyrus (PoCG). The second pattern was characterized by HMS activity with ictal spread to middle cingulate cortex (MCC) and posterior cingulate cortex (PCC).ConclusionAura type (e.g., body image disturbance and vestibular response), BATS, and HMS are the main indicators of precuneal epilepsy. Scalp EEG is of little use when attempting to localize precuneal seizures. Our findings indicate that the clinical characteristics of precuneal epilepsy vary among patients, and that the final electro–clinical phenotype depends on the pattern of seizure spread.

EEG signal classification using universum support vector machine

Support vector machine (SVM) has been used widely for classification of electroencephalogram (EEG) signals for the diagnosis of neurological disorders such as epilepsy and sleep disorders. SVM shows good generalization performance for high dimensional data due to its convex optimization problem. The incorporation of prior knowledge about the data leads to a better optimized classifier. Different types of EEG signals provide information about the distribution of EEG data. To include prior information in the classification of EEG signals, we propose a novel machine learning approach based on universum support vector machine (USVM) for classification. In our approach, the universum data points are generated by selecting universum from the EEG dataset itself which are the interictal EEG signals. This removes the effect of outliers on the generation of universum data. Further, to reduce the computation time, we use our approach of universum selection with universum twin support vector machine (UTSVM) which has less computational cost in comparison to traditional SVM. For checking the validity of our proposed methods, we use various feature extraction techniques for different datasets consisting of healthy and seizure signals. Several numerical experiments are performed on the generated datasets and the results of our proposed approach are compared with other baseline methods. Our proposed USVM and proposed UTSVM show better generalization performance compared to SVM, USVM, Twin SVM (TWSVM) and UTSVM. The proposed UTSVM has achieved highest classification accuracy of 99% for the healthy and seizure EEG signals.

Hardware design of multiclass SVM classification for epilepsy and epileptic seizure detection

An automatic detection system for distinguishing healthy, ictal, and inter-ictal EEG signals plays an important role in medical practice. This paper presents a very large scale integration (VLSI) architecture of three-class classification for epilepsy and seizure detection. In order to find out the most efficient three-class classification scheme for hardware implementation, several multiclass non-linear support vector machine (NLSVM) classifiers are compared and validated using software implementation. Finally, the one-against-one (OAO) multiclass NLSVM is selected due to its highest accuracy. The designed system consists of a discrete wavelet transform (DWT)-based feature extraction module, a modified sequential minimal optimization (MSMO) training module, and an OAO multiclass classification module. A lifting structure of Daubechies order 4 wavelet is introduced in three-level DWT to save circuit area and speed up the computational time. The MSMO is used for on-chip training. The circuit of the largest absolute value decision is designed to avoid the unclassifiable problem in the OAO multiclass classification. The designed system is implemented on a field-programmable gate array (FPGA) platform and evaluated using the publicly available epilepsy dataset. The experimental results demonstrate that the designed system achieves high accuracy with low-dimensional feature vectors.